Chaos

def chaos():

  """
  Generate an inspirational sentence.
    
  Ensure that you are in the proper state of mind before running. ॐ
  """
    
  # Makes a sentence skeleton by randomnly choosing sentence structures
  # Reconstructs the sentence by sampling from corresponding POS 
  # Returns the reconstituted sentence

Output

Issues

• Possessives+contractions don’t play nicely with NLTK
• Assuming every word is equally likely to occur is dumb

N-gram Model Sentence Generator

def ngram_sentence(num_words, model):

    # Pads the sentence with start/end tokens
    # Predict the next token given the previous n-1 tokens
    # All of the observed n-grams in our corpus represent a prob. dist.
    # Simulate that probability distribution
    # Use probabilities to build a sentence using n-grams 
    # Returns a cleaner sentence (first word capitalized, periods, etc)

Review:

A bigram makes a prediction for a word based on the one before, and a trigram makes a prediction for the word based on the two words before that.

Duality

def duality():   

    """
    You must only concentrate on the next step, the next breath, 
    the next stroke of the broom, and the next, and the next. 
    Nothing else. ॐ
    
    (Bigram Model)
    """    
    
    # Returns an inspirational sentence made with a bigram model

Output

Issues

• Bigrams were a significant improvement over V0.
• But, while you’re doing bigrams, might as well do trigrams.

Third Eye

def third_eye():

    """
    Three things cannot long be hidden: the sun, the moon, 
    and the truth. ॐ
    
    (Trigram Model)
    """
    
    # Returns an inspirational sentence made with trigrams

Output

Output

• Subject/verb agreement
• Perfect an RNN (better than the LSTM we tried).

Instagram account